[gentoo-dev] The Dreaded herd tag

[gentoo-dev] The Dreaded herd tag

[Alec Warner]

I request that this tag be made optional in the metadata.xml DTD.

While ideally it is beneficial to have every package in a herd, in 
practice this doesn't occur.

22:28 <@omp> $ herdstat -pq no-herd | wc -l
22:28 <@omp> 1819
22:28 <@omp> looks like a lot of fixing is needed :)

nearly 1/5 of our tree is herdless.

Is a "real herd" a real requirement of a package?  I would say 
realistically no.  Thus the herd tag should be optional but highly 
encouraged.


Why do I care?

Because people don't always put "no-herd", they can misspell it, they 
put non-standard stuff in the herd tag, they put bug-wranglers in the 
herd tag...etc...

It's difficult to search for packages without a herd when 'no-herd' is 
really N different strings.

If your package is not in a herd, just leave the tag out.  We can adjust 
the searching tools to list packages with no herd tag.

-Alec Warner
antarus@gentoo.org
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] The Dreaded herd tag

[Mike Frysinger]

On 10/27/06, Alec Warner <antarus@gentoo.org> wrote:
> Because people don't always put "no-herd", they can misspell it, they
> put non-standard stuff in the herd tag, they put bug-wranglers in the
> herd tag...etc...

this is kind of a lame reason ... there is no real reason we cant add
<herd> tag validation to repoman

if it isnt in this file:
http://sources.gentoo.org/gentoo/misc/herds.xml
then it isnt a valid herd
-mike
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] The Dreaded herd tag

[Doug Goldstein]

[-- Attachment #1: Type: text/plain, Size: 656 bytes --]

Mike Frysinger wrote:
> On 10/27/06, Alec Warner <antarus@gentoo.org> wrote:
>> Because people don't always put "no-herd", they can misspell it, they
>> put non-standard stuff in the herd tag, they put bug-wranglers in the
>> herd tag...etc...
> 
> this is kind of a lame reason ... there is no real reason we cant add
> <herd> tag validation to repoman
> 
> if it isnt in this file:
> http://sources.gentoo.org/gentoo/misc/herds.xml
> then it isnt a valid herd
> -mike

Wanna add "no-herd" to that list then? Cause right now there's no way to
 identify no herd.

-- 
Doug Goldstein <cardoe@gentoo.org>
http://dev.gentoo.org/~cardoe/


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 252 bytes --]

Re: [gentoo-dev] The Dreaded herd tag

[George Shapovalov]

субота, 28. жовтень 2006 08:01, Doug Goldstein Ви написали:
> Wanna add "no-herd" to that list then? Cause right now there's no way to
>  identify no herd.
And there should not be. This entry is *invalid* and *disallowed*. Every 
ebuild must belong to a valid herd. Period.

I am afraid, that policy may only be inferred by reading the metadata page of 
dev handbook:
http://www.gentoo.org/proj/en/devrel/handbook/handbook.xml?part=2&chap=4
Sadly, this is one of the few things that did not get fomalized properly since 
the "oral tradition" times. Should we formalize it now?

If we do it any other way, we cannot rely on the metadata for the purposes of 
tracking unmaintained ebuilds. Yes, some ebuilds are not herded right now. 
Wanna guess how many of those happen to be stale? Then, <20% unherded 
packages? - I consider this a very good result! Remember, there is 
no "absolute correctness", so I would say right now the idea works for all 
practical purposes.

George

-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] The Dreaded herd tag

[Chris Gianelloni]

[-- Attachment #1: Type: text/plain, Size: 693 bytes --]

On Sat, 2006-10-28 at 08:28 +0200, George Shapovalov wrote:
> Wanna guess how many of those happen to be stale?

I would suspect fewer than you think.  As an example, I have a few
packages which belong to no herd, but have me listed as maintainer.
Many of the no-herd packages are the same.  Not being grouped with other
packages doesn't mean it is unmaintained.

So this bears the question, what is the proper solution?

Make <herd> optional?
Force the maintainer's email into <herd> for packages without a herd?

-- 
Chris Gianelloni
Release Engineering Strategic Lead
Alpha/AMD64/x86 Architecture Teams
Games Developer/Council Member/Foundation Trustee
Gentoo Foundation

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] The Dreaded herd tag

[Jim Ramsay]

[-- Attachment #1: Type: text/plain, Size: 946 bytes --]

On Mon, Oct 30, 2006 at 11:09:53AM -0500, Chris Gianelloni wrote:
> On Sat, 2006-10-28 at 08:28 +0200, George Shapovalov wrote:
> > Wanna guess how many of those happen to be stale?
> 
> I would suspect fewer than you think.  As an example, I have a few
> packages which belong to no herd, but have me listed as maintainer.
> Many of the no-herd packages are the same.  Not being grouped with other
> packages doesn't mean it is unmaintained.

I agree.  All my rox stuff (rox-base/* and rox-extra/*) kind of
fits into many different herds.  Or no herd.  Or maybe its own
new herd.

> So this bears the question, what is the proper solution?
> 
> Make <herd> optional?
> Force the maintainer's email into <herd> for packages without a herd?

Or force each maintainer of herdless packages to create their own
special herd.  Which I may end up doing with my rox packages
anyway.

-- 
Jim Ramsay
Gentoo/Linux Developer (rox)

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] The Dreaded herd tag

[David Shakaryan]

[-- Attachment #1: Type: text/plain, Size: 938 bytes --]

Alec Warner wrote:
> I request that this tag be made optional in the metadata.xml DTD.

++

In my opinion, an empty tag or one with "no-herd" is rather silly when
the requirement for the tag can just be removed.

The alternative is, of course, requiring every package to belong to a
real herd. Although this is not impossible, it would be quite hard to
implement as quite a few packages do not really fit into any existing
category, as you mentioned in #-dev. Also, someone will have to go
through the ~2000 packages and figure out what herd they belong to,
which seems like a strenuous job.

I say the first is a much more feasible solution. I'll even be glad to
help in removing the herd tags for the affected ebuilds if this is the
path chosen.

I say we act on this now, as the further off we put it until, the more
metadata.xml files there will be to fix.

-- 
David Shakaryan
GnuPG Public Key: 0x4B8FE14B


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] The Dreaded herd tag

[George Shapovalov]

субота, 28. жовтень 2006 05:23, David Shakaryan Ви написали:
> Alec Warner wrote:
> > I request that this tag be made optional in the metadata.xml DTD.
>
> ++
>
> In my opinion, an empty tag or one with "no-herd" is rather silly when
> the requirement for the tag can just be removed.
Which is exactly why these are disallowed. Or at least that was the original 
intention, which (unfortunately) was not enforced strong enough. But then, 
given that we started with *no herds at all*, I don't see how it would be 
possible to realistically enforce from the beginning. Now it looks like we 
are actually strating to "get there". Besides, there is no "no-herd" tag, no 
matter what excuses people putting it in the metadata come up with. 

> The alternative is, of course, requiring every package to belong to a
> real herd. 
Which is exactly the case. This is the policy, as it was set from the 
introduction of herds. Unfortunately, as I said, because it was technically 
infeasible right from the beginning, it was not vigorously enforced, but, as 
Mike, mentioned, there is no reason we cannot finally start doing this now.

> Although this is not impossible, it would be quite hard to 
> implement as quite a few packages do not really fit into any existing
> category, as you mentioned in #-dev. Also, someone will have to go
> through the ~2000 packages and figure out what herd they belong to,
> which seems like a strenuous job.
What is it now < 20%? Phew. I am sure treecleaners would tell you that their 
estimate of "stale" ebuilds in portage goes over that percentage ;). Besides, 
in the beginning unherded ebuilds were at 100%.

One of the reasons herds were introduced was to explicitly see what packages 
lack maintenance. It is possible for the ebuild to be in the herd, but be 
supported by the developer not on the herd. See the <role> tag. Also, there 
can be one-dev herds.

George

-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] The Dreaded herd tag

[Robin H. Johnson]

[-- Attachment #1: Type: text/plain, Size: 1278 bytes --]

On Sat, Oct 28, 2006 at 08:11:37AM +0200, George Shapovalov wrote:
> One of the reasons herds were introduced was to explicitly see what packages 
> lack maintenance. It is possible for the ebuild to be in the herd, but be 
> supported by the developer not on the herd. See the <role> tag. Also, there 
> can be one-dev herds.
I have a number of specialized packages that I maintain, such as
sys-block/qla-fc-firmware, that cannot be classified as any existing
herd, and are specialized enough inventing a new herd for them would not
really help.

The point of herds as I saw it originally, was to capture packages that
do NOT have any explicit maintainer. 
	
I've also lost bugs in the past on my packages where I'm not in the
herd, but I am the actual maintainer, because the bug was assigned to
the herd alias, and I didn't see it until several months later, when
somebody finally asked me directly, or reassigned it to me.

For no-herd, I say we should add it to the valid herds list, and
validate all metadata files, and require that if no-herd is used, an
explicit maintainer is present (using the active list of developers).

-- 
Robin Hugh Johnson
E-Mail     : robbat2@gentoo.org
GnuPG FP   : 11AC BA4F 4778 E3F6 E4ED  F38E B27B 944E 3488 4E85

[-- Attachment #2: Type: application/pgp-signature, Size: 232 bytes --]

Re: [gentoo-dev] The Dreaded herd tag

[Mike Frysinger]

[-- Attachment #1: Type: text/plain, Size: 1121 bytes --]

On Saturday 28 October 2006 02:46, Robin H. Johnson wrote:
> On Sat, Oct 28, 2006 at 08:11:37AM +0200, George Shapovalov wrote:
> > One of the reasons herds were introduced was to explicitly see what
> > packages lack maintenance. It is possible for the ebuild to be in the
> > herd, but be supported by the developer not on the herd. See the <role>
> > tag. Also, there can be one-dev herds.
>
> I have a number of specialized packages that I maintain, such as
> sys-block/qla-fc-firmware, that cannot be classified as any existing
> herd, and are specialized enough inventing a new herd for them would not
> really help.

declaring no herd for maintainership here makes sense ... requiring a <herd> 
tag and forcing it to "no-herd" keeps things explicit ...

on the topic of finding unmaintained packages:
if there is no herd and no maintainer, should we just cut metadata.xml ?  or 
do we recommend people to stick in <herd>no-herd</herd> ?  the former would 
help with people sticking in bogus things like a maintainer of bug-wranglers 
(really maintainer-needed would make more sense) ...
-mike

[-- Attachment #2: Type: application/pgp-signature, Size: 827 bytes --]

Re: [gentoo-dev] The Dreaded herd tag

[Chris Gianelloni]

[-- Attachment #1: Type: text/plain, Size: 1574 bytes --]

On Sat, 2006-10-28 at 03:05 -0400, Mike Frysinger wrote:
> On Saturday 28 October 2006 02:46, Robin H. Johnson wrote:
> > On Sat, Oct 28, 2006 at 08:11:37AM +0200, George Shapovalov wrote:
> > > One of the reasons herds were introduced was to explicitly see what
> > > packages lack maintenance. It is possible for the ebuild to be in the
> > > herd, but be supported by the developer not on the herd. See the <role>
> > > tag. Also, there can be one-dev herds.
> >
> > I have a number of specialized packages that I maintain, such as
> > sys-block/qla-fc-firmware, that cannot be classified as any existing
> > herd, and are specialized enough inventing a new herd for them would not
> > really help.
> 
> declaring no herd for maintainership here makes sense ... requiring a <herd> 
> tag and forcing it to "no-herd" keeps things explicit ...

That's what I think is best.

> on the topic of finding unmaintained packages:
> if there is no herd and no maintainer, should we just cut metadata.xml ?  or 
> do we recommend people to stick in <herd>no-herd</herd> ?  the former would 
> help with people sticking in bogus things like a maintainer of bug-wranglers 
> (really maintainer-needed would make more sense) ...

Well, we enforce the maintainer tag if herd is no-herd.  Then, we only
allow valid devs, and maintainer-needed in maintainer.  That should
resolve the problem.

-- 
Chris Gianelloni
Release Engineering Strategic Lead
Alpha/AMD64/x86 Architecture Teams
Games Developer/Council Member/Foundation Trustee
Gentoo Foundation

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] The Dreaded herd tag

[George Shapovalov]

понеділок, 30. жовтень 2006 17:16, Chris Gianelloni Ви написали:
> Well, we enforce the maintainer tag if herd is no-herd.  Then, we only
With this explicit requirement I think it should be Ok.

> allow valid devs, and maintainer-needed in maintainer.  
Should we also disallow adding new no-herd/maintainer-needed ebuilds?
(As the apparent use of maintainer-needed is to track the ebuilds already in 
the tree that need some love).

George

-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] The Dreaded herd tag

[Olivier Crete]

On Mon, 2006-30-10 at 17:40 +0100, George Shapovalov wrote:
> понеділок, 30. жовтень 2006 17:16, Chris Gianelloni Ви написали:
> > allow valid devs, and maintainer-needed in maintainer.  
> Should we also disallow adding new no-herd/maintainer-needed ebuilds?
> (As the apparent use of maintainer-needed is to track the ebuilds already in 
> the tree that need some love).

Isn't adding an ebuild without setting oneself or one's herd as the
maintainer already forbidden?

-- 
Olivier Crête
tester@gentoo.org
Gentoo Developer


-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] The Dreaded herd tag

[George Shapovalov]

понеділок, 30. жовтень 2006 17:49, Olivier Crete Ви написали:
> > Should we also disallow adding new no-herd/maintainer-needed ebuilds?
> > (As the apparent use of maintainer-needed is to track the ebuilds already
> > in the tree that need some love).
>
> Isn't adding an ebuild without setting oneself or one's herd as the
> maintainer already forbidden?
Yes, but since we are talking now about changing rules I am trying to get as 
much spelled out as makes sense. Sorry, just being too pedantic here :).

George

-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] The Dreaded herd tag

[Paul de Vrieze]

[-- Attachment #1: Type: text/plain, Size: 1351 bytes --]

On Saturday 28 October 2006 08:11, George Shapovalov wrote:

> Which is exactly why these are disallowed. Or at least that was the
> original intention, which (unfortunately) was not enforced strong enough.
> But then, given that we started with *no herds at all*, I don't see how it
> would be possible to realistically enforce from the beginning. Now it looks
> like we are actually strating to "get there". Besides, there is no
> "no-herd" tag, no matter what excuses people putting it in the metadata
> come up with.

Being the one who came up with the no-herd tag I'd like to explain things a 
bit. Basically when we started there were no herds and packages didn't belong 
to them. It was agreed that every package should be put in a herd, but also 
that metadata.xml files were to be added and their existence enforced by 
repoman. This enforcing was easy so it happened before it could be expected 
that all maintained packages (they needed metadata.xml files) could have 
found themselves a herd. Then I thought it is better to temporarilly allow 
adding no-herd than to have everyone come up with his own version of the 
same. I should have remembered that there is nothing as permanent as a 
temporary solution.

Paul

-- 
Paul de Vrieze
Gentoo Developer
Mail: pauldv@gentoo.org
Homepage: http://www.devrieze.net

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: [gentoo-dev] The (lack of) use of herds

[Marius Mauch]

On Fri, 27 Oct 2006 22:46:30 -0400
Alec Warner <antarus@gentoo.org> wrote:

> I request that this tag be made optional in the metadata.xml DTD.
> 
> While ideally it is beneficial to have every package in a herd, in 
> practice this doesn't occur.
> 
> 22:28 <@omp> $ herdstat -pq no-herd | wc -l
> 22:28 <@omp> 1819
> 22:28 <@omp> looks like a lot of fixing is needed :)
> 
> nearly 1/5 of our tree is herdless.
> 
> Is a "real herd" a real requirement of a package?  I would say 
> realistically no.  Thus the herd tag should be optional but highly 
> encouraged.

Well, I'd go further and question the whole herd concept. What benefits
do we actually gain by having "herds"? For the most part it's just a
way to associate a package with a mail alias, but for that I don't
really see the need for this layer of indirection. It actually creates
problems by itself as the herd data ("members" in herds.xml) gets out
of sync with the mail data (alias members), then there is the (mostly
historical) issue of having two copies of the same file getting out of
sync, the permanent confusion of herds, herd maintainers and projects,
and the problem just shown by Alec.
So are there any other benefits in having herds as opposed to just
adding a <maintainer><email>mail-alias@gentoo.org</email></maintainer>
element to metadata.xml and getting rid of the complete herds concept?

Marius

-- 
Public Key at http://www.genone.de/info/gpg-key.pub

In the beginning, there was nothing. And God said, 'Let there be
Light.' And there was still nothing, but you could see a bit better.
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] The (lack of) use of herds

[Mike Frysinger]

[-- Attachment #1: Type: text/plain, Size: 1011 bytes --]

On Saturday 28 October 2006 22:43, Marius Mauch wrote:
> Well, I'd go further and question the whole herd concept. What benefits
> do we actually gain by having "herds"? For the most part it's just a
> way to associate a package with a mail alias, but for that I don't
> really see the need for this layer of indirection. It actually creates
> problems by itself as the herd data ("members" in herds.xml) gets out
> of sync with the mail data (alias members), then there is the (mostly
> historical) issue of having two copies of the same file getting out of
> sync, the permanent confusion of herds, herd maintainers and projects,
> and the problem just shown by Alec.
> So are there any other benefits in having herds as opposed to just
> adding a <maintainer><email>mail-alias@gentoo.org</email></maintainer>
> element to metadata.xml and getting rid of the complete herds concept?

just because you're on the alias doesnt mean you want to be responsible for 
the packages in the herd
-mike

[-- Attachment #2: Type: application/pgp-signature, Size: 827 bytes --]

Re: [gentoo-dev] The (lack of) use of herds

[Marius Mauch]

On Sat, 28 Oct 2006 23:00:04 -0400
Mike Frysinger <vapier@gentoo.org> wrote:

> On Saturday 28 October 2006 22:43, Marius Mauch wrote:
> > Well, I'd go further and question the whole herd concept. What
> > benefits do we actually gain by having "herds"? For the most part
> > it's just a way to associate a package with a mail alias, but for
> > that I don't really see the need for this layer of indirection. It
> > actually creates problems by itself as the herd data ("members" in
> > herds.xml) gets out of sync with the mail data (alias members),
> > then there is the (mostly historical) issue of having two copies of
> > the same file getting out of sync, the permanent confusion of
> > herds, herd maintainers and projects, and the problem just shown by
> > Alec. So are there any other benefits in having herds as opposed to
> > just adding a
> > <maintainer><email>mail-alias@gentoo.org</email></maintainer>
> > element to metadata.xml and getting rid of the complete herds
> > concept?
> 
> just because you're on the alias doesnt mean you want to be
> responsible for the packages in the herd

So if now you're on a herd alias but not listed in herds.xml for a herd,
what practical difference does it make if the herd exists or not?

Marius

-- 
Public Key at http://www.genone.de/info/gpg-key.pub

In the beginning, there was nothing. And God said, 'Let there be
Light.' And there was still nothing, but you could see a bit better.
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] The (lack of) use of herds

[Mike Frysinger]

[-- Attachment #1: Type: text/plain, Size: 209 bytes --]

On Sunday 29 October 2006 01:26, Marius Mauch wrote:
> So if now you're on a herd alias but not listed in herds.xml for a herd,
> what practical difference does it make if the herd exists or not?

huh ?
-mike

[-- Attachment #2: Type: application/pgp-signature, Size: 827 bytes --]

Re: [gentoo-dev] The (lack of) use of herds

[Alec Warner]

Mike Frysinger wrote:
> On Sunday 29 October 2006 01:26, Marius Mauch wrote:
>> So if now you're on a herd alias but not listed in herds.xml for a herd,
>> what practical difference does it make if the herd exists or not?
> 
> huh ?
> -mike

I think one of his points is that you still get the bugmail...I guess?

I'm on the python team alias but not in the herd...I don't want have 
people ask me to fix python stuff because I dunno how to do it ;) 
However I like knowing 'what is going on in python town'.

Peace out homies.
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] The (lack of) use of herds

[Richard Fish]

On 10/28/06, Marius Mauch <genone@gentoo.org> wrote:
> Well, I'd go further and question the whole herd concept.

It also gives users the impression that there is an entire "team" of
people maintaining a package,when in fact it might be just one or two
people.

-Richard
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] The (lack of) use of herds

[Elfyn McBratney]

On 29/10/06, Richard Fish <bigfish@asmallpond.org> wrote:
> On 10/28/06, Marius Mauch <genone@gentoo.org> wrote:
> > Well, I'd go further and question the whole herd concept.
>
> It also gives users the impression that there is an entire "team" of
> people maintaining a package,when in fact it might be just one or two
> people.

A single person doesn't constitute a team [1].  More than one person does...

Best,
Elfyn

[1] Unless you're SpanKY.
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] The (lack of) use of herds

[Roy Marples]

On Monday 30 October 2006 10:26, Elfyn McBratney wrote:
> A single person doesn't constitute a team [1].  More than one person
> does...
>
> [1] Unless you're SpanKY.

You forget that vapier also works with SpanKY ;)

-- 
Roy Marples <uberlord@gentoo.org>
Gentoo/Linux Developer (baselayout, networking)
-- 
gentoo-dev@gentoo.org mailing list